Hybrid partitioning has been recognized as a technique to achieve query optimization in relational and object -oriented databases. Due to the increasing availability of multimedia applications, there is an interest in using partitioning techniques in multimedia da tabases in order to take advantage of the reduction in the number of pages required to answer a query and to minimize data exchange among sites. Never theless, until now only vertical and horizontal partitioning have been used in multimedia databases. This paper presents a hybrid partitioning method for multimedia databases. This method takes into account the size of the attributes and the selectivity of the predicates in order to generate hybrid partitioning schemes that reduce the execution cost of the que ries. A cost model for evaluating hybrid partitioning schemes in distributed multimedia databases was developed. Experiments in a multimedia database benchmark were performed in order to demonstrate the efficiency of our ap proach.